2
$\begingroup$

I'm trying to compute the integral $$\int_{a}^{b}(\sin( 1 + x ) + \cos( 1 + x ))e^{-x}\ dx$$

using importance sample Monte Carlo method. The exercise ask to use Cauchy Distribution to resolve the integral. Then $$\frac{1}{n}\sum_{i}^{n}\frac{f(x_i)}{g(x_{i})}$$ is an aproximation to the integral, where $g(x) = \frac{1}{\pi(1+x^2)}$(Cauchy distribution). To generate random numbers in a range $[a, b]$ following Cauchy distribution I used the inverse transformation: $$X = \tan(\arctan(a) + U(\arctan(b)-\arctan(a)))\quad \Rightarrow\quad \ U \sim(0,1)$$

but I did not get success.

$\endgroup$
3
$\begingroup$

$\newcommand{\+}{^{\dagger}} \newcommand{\angles}[1]{\left\langle\, #1 \,\right\rangle} \newcommand{\braces}[1]{\left\lbrace\, #1 \,\right\rbrace} \newcommand{\bracks}[1]{\left\lbrack\, #1 \,\right\rbrack} \newcommand{\ceil}[1]{\,\left\lceil\, #1 \,\right\rceil\,} \newcommand{\dd}{{\rm d}} \newcommand{\down}{\downarrow} \newcommand{\ds}[1]{\displaystyle{#1}} \newcommand{\expo}[1]{\,{\rm e}^{#1}\,} \newcommand{\fermi}{\,{\rm f}} \newcommand{\floor}[1]{\,\left\lfloor #1 \right\rfloor\,} \newcommand{\half}{{1 \over 2}} \newcommand{\ic}{{\rm i}} \newcommand{\iff}{\Longleftrightarrow} \newcommand{\imp}{\Longrightarrow} \newcommand{\isdiv}{\,\left.\right\vert\,} \newcommand{\ket}[1]{\left\vert #1\right\rangle} \newcommand{\ol}[1]{\overline{#1}} \newcommand{\pars}[1]{\left(\, #1 \,\right)} \newcommand{\partiald}[3][]{\frac{\partial^{#1} #2}{\partial #3^{#1}}} \newcommand{\pp}{{\cal P}} \newcommand{\root}[2][]{\,\sqrt[#1]{\vphantom{\large A}\,#2\,}\,} \newcommand{\sech}{\,{\rm sech}} \newcommand{\sgn}{\,{\rm sgn}} \newcommand{\totald}[3][]{\frac{{\rm d}^{#1} #2}{{\rm d} #3^{#1}}} \newcommand{\ul}[1]{\underline{#1}} \newcommand{\verts}[1]{\left\vert\, #1 \,\right\vert} \newcommand{\wt}[1]{\widetilde{#1}}$ The probability distribution which is normalized in $\ds{\left[a,b\right)}$ is given by: $$ {\rm P}\pars{x} = {1 \over \arctan\pars{b} - \arctan\pars{a}}\,{1 \over x^{2} + 1} $$

$$ \int_{a}^{x}{\rm P}\pars{\xi}\,\dd\xi = \int_{0}^{U}\dd U'\quad\imp\quad {\arctan\pars{X} - \arctan\pars{a} \over \arctan\pars{b} - \arctan\pars{a}} =U $$ which leads to $\ds{\arctan\pars{X} =\arctan\pars{a} + \bracks{\arctan\pars{b} - \arctan\pars{a}}U}$: \begin{align} X = \tan\pars{\arctan\pars{a} + \bracks{\arctan\pars{b} - \arctan\pars{a}}} ={a + \tan\pars{\bracks{\arctan\pars{b} - \arctan\pars{a}}U} \over 1 - a\tan\pars{\bracks{\arctan\pars{b} - \arctan\pars{a}}U}}\tag{1} \end{align}

$$ \sum_{i = 1}^{N}\fermi\pars{x_{i}} \approx N\int_{a}^{b}{\rm P}\pars{x}\fermi\pars{x}\,\dd x $$

$$ \int_{a}^{b}{\fermi\pars{x} \over x^{2} + 1} \approx {\arctan\pars{b} - \arctan\pars{a} \over N}\sum_{i = 1}^{N}\fermi\pars{x_{i}} $$

\begin{align} &\color{#000}{\large% \int_{a}^{b}\bracks{\sin\pars{1 + x} + \cos\pars{1 + x}}\expo{-x}\,\dd x} \\[3mm]&\approx\color{#00f}{\large% {\arctan\pars{b} - \arctan\pars{a} \over N}\times} \\[3mm]&\color{#00f}{\large\sum_{i = 1}^{N} \bracks{\sin\pars{1 + x_{i}} + \cos\pars{1 + x_{i}}}\expo{-x_{i}} \pars{x_{i}^{2} + 1}} \end{align} with $\ds{\braces{x_{i}}}$ generated by $\pars{1}$ and associated $\ds{\braces{U_{i}}}$ generated by an uniform distribution in $\ds{\left[0,1\right)}$.

$\endgroup$
  • $\begingroup$ thanks very much for the help. For example if I wanted to use the exponential distribution, it would be normalized in [a, b] is given by: $P(x) = \frac{e^{-x + a}}{e^{-b} - e^{-a}}$.Sorry about that more this question is that I'm comparing the distributions $\endgroup$ – user50098 May 11 '14 at 23:17
  • $\begingroup$ @user50098 You're welcome. Sorry it's quite late. $\endgroup$ – Felix Marin Nov 16 '16 at 6:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.